Download ClusterStor A200

ClusterStor A200
Delivers Industry-best cost, performance, and availability metrics
The ClusterStor A200 is a tiered active archive object store for the
ClusterStor product line. The ClusterStor A200 solution includes active
archiving and hierarchical storage management components that
provides an automatic policy-driven archive storage tier for ClusterStor
systems.
LOWEST
COST
HIGHEST
PERFORMANCE
The ClusterStor A200 solution is engineered for productivity and
provides an optimal combination of cost, performance, and availability
for the most demanding HPC and commercial workloads. The A200
tiered archive delivers up to 50% lower cost than HPC primary storage
‘

and 2-3x more performance and higher availability than other archive
solutions in the marketplace, while providing near limitless scale-out
capabilities in terms of capacity, number of objects and throughput.
The ClusterStor A200 utilizes network erasure coding that is tuned to
the storage platform to deliver unmatched availability, scalability and
performance for active archiving applications. The A200 is capable of
d
providing extremely high levels of durability by reconstructing data from
a failed 8TB drive in less than 1 hour in the reference configuration.
The A200 utilizes massive parallelization to deliver linear performance
and capacity scalability to meet the active archiving needs of the most
MAXIMUM
RELIABILITY
demanding high performance computing environments. The
ClusterStor A200 has been designed with extreme system resiliency,
with no single points of failure and the ability to replace any failed
component without downtime.
Benefits
Customers benefit from industry-leading economics, not only in terms
of storage and performance costs but also from reduced data center
floor space, power, cooling and administrative costs; up to 60% better
than competitive active archive solutions. Customers using the A200
can scale the system indefinitely and save up to 50% in storage and
operational costs by migrating static data from Lustre or Spectrum
Scale primary storage to the A200 active archive object store.
Storage Value-Utilizes cost effective SMR drives & efficient erasure
coding.
Economics
Up to 50% $/TB storage cost savings vs HPC primary storage
Limitless Scale - 3.4 Undecillion (2^128) objects, unlimited object
size, unlimited storage capacity.
Utilize most cost effective drives - 8TB+ SMR SATA HDDs
Linear performance scaling - Up to 10GB/s per rack, add additional
Cost efficient 8+2 erasure coding
racks to increase performance linearly (up to the network limit).
Performance
Parallel read/write access paths
4 "Nines" of Availability. Less than 55 min of downtime per year for
base configuration.
Linear multi-session scaling increases research productivity
Read/write throughput of up to10GB/s per rack (7SSUs)
Availability/Durability
4 “nines” availability for base configuration
No single points of failure, replace any failed component without downtime.
Improved data availability & durability via network erasure coding
Features
Archive object storage tier for ClusterStor platform.
Combine with ClusterStor HSM to provide automatic policy-driven
data migration & retrieval.
Optimized for a balance of cost, performance & density.
Utilizes network erasure coding to provide high levels of data
availability and data durability.
No single points of failure, replace any failed component without
downtime.
Dual 10Gb Ethernet node connectivity.
ClusterStor A200
Delivers Industry-best cost, performance, and availability metrics
ClusterStor A200 Tiered Archive
The ClusterStor A200 is a pre-configured rack-based storage solution
that delivers predictable performance and eliminates single points of
failure. The A200 consists of 5-7 scalable storage units (SSU), a
dedicated ClusterStor management unit (CMU) and redundant
Switch A
Switch B
10/40GbE switches.
Mgmt Server
A200 Scalable Storage unit
The Scalable Storage Unit (SSU) is the storage sub-system for the
A200 active archive. Each 5U84 A200 SSU contains an embedded
SSU7
storage controller, eighty two 8TB SMR drives, two metadata drives,
dual 10GigE network ports, and provides 524 usable TB per tray (over
3.6 usable PB per rack). The result is an easy to deploy, easy to use
SSU6
and easy to manage solution. There is no need to guess at how to
scale performance for multi-session workloads; each SSU is a
balanced performance building block delivering a predictable level of
SSU5
performance and system can be scaled by simply adding SSUs and
expansion racks as needed.
SSU4
Additional SSU's and expansion racks can non-disruptively be added
as needed to provide linear capacity and performance scalability.
SSU3
Overall system performance is directly proportional to the number of
SSUs to satisfy performance and/or data capacity needs.
SSU2
Seagate Capacity Optimized 8TB SMR Hard
Drives
SSU1
Shingled Magnetic Recording (SMR) drives are optimized for cost,
efficiency and capacity as data is written wide and the next adjacent
track write “trims” the previous track. This makes SMR drives a perfect
choice for archival applications, where capacity and cost are
A200 Base Rack
paramount. Readable tracks are narrower than originally written tracks,
which results in 30-40% more tracks being stored on the platter than
standard Perpendicular Magnetic Recording (current SAS technology)
drives.
A200 ClusterStor Manager
The A200 ClusterStor Manager is a comprehensive system
management application that is part of the CS A200's distributed
management framework and is responsible for consolidating health
and performance monitoring and administration in a single managed
system view.
ClusterStor Manager consolidates management of the CS A200
storage infrastructure, network erasure coding data protection layer,
and HSM management and monitoring into a single, easy-to-use
administrator interface.
Initial installation, creation of data layouts, ongoing monitoring and
management, expansion and software upgrades are accessed via a
single pane of glass. This brings unprecedented system visibility to the
storage management system administrator.
ClusterStor A200
Delivers Industry-best cost, performance, and availability metrics
CS A200 Data Protection
HSM Data Mover
Customer Application
Archive Software Library
Archive Software Library
40GbE Networks
10GbE Networks
8 Data Chunks
2 Parity Chunks
ClusterStor A200 Data Layouts
The CS A200 system supports network erasure coding to provide data
and disaster protection. Multiple network erasure coding schemas
(called “Data Layouts”) can be concurrently defined and utilized in a
single A200 system to deliver separation of data across SSUs if
desired.
Seagate’s network erasure coding distributes the data+parity shards
across multiple SSUs which enables data to be available even if
multiple SSUs or individual drives fail concurrently. The default
supported CS A200 erasure coding schema is 8+2, meaning that
objects are sharded into eight data shards and 2 parity chunks and
written across 10 networked SSU’s. An A200 configuration containing
ten or more SSUs maintains full data read availability even in the event
of two concurrent SSU failures or two concurrent drive failures. In
addition, network erasure coding extends the number of drives that
can participate in data reconstruction across SSUs, allowing for
massively increased parallelism and reconstruction speed. Increased
reconstruction speed in turn shortens the failed drive data
reconstruction time, leading to an increase in data durability as the
system scales.
ClusterStor
ClusterStor HSM
Key Storage Tiering (HSM) Capabilities
ClusterStor HSM is designed to offload data from a higher level
storage tier to a less expensive ClusterStor A200 Active Archive
storage tier based on customer defined policies to meet specific cost,
performance and quality of service requirements while still maintaining
transparent access to data. The initial ClusterStor HSM release is
designed to manage Lustre ClusterStor configurations, and support for
IBM Spectrum Scale (GPFS) is planned for a follow-on release.
ClusterStor HSM
•
•
•
Policy Engine
Database
Data Movers
ClusterStor HSM Benefits
Storage costs – Up to 50% cost reduction by moving data off of
expensive tier 1 storage.
Data availability & durability – Improve availability & durability of
archived data for long-term retention.
Performance – Reduce primary storage fragmentation & latency /
seek times.
Automation / User productivity – Automation reduces effort & setup
time to run a job.
ClusterStor Lustre Active Archive Use Case
CS A200
ClusterStor HSM
ClusterStor HSM is a gateway appliance that utilizes data movers that
act as Lustre clients to transfer files between ClusterStor primary
storage Object Storage Targets(OSTs) and the ClusterStor A200
storage target based on customer-defined policies. Customer-specified
policies determine when data is copied (migrated) to the secondary
storage target, and when migrated files are stubbed in primary storage.
Files are typically stubbed based on last access time, size, creation
date, and/or some combination of other file properties. Migrated files
can be also stubbed on demand when primary storage free space falls
below a specific threshold. When a Lustre client/user attempts to use
this file, it is automatically and transparently retrieved from the
secondary storage tier before proceeding. Once a file is re-hydrated, it
can again be stubbed per policy once it meets policy criteria.
ClusterStor A200
ClusterStor A200 Specifications - General Information (1) and (2)
System Capacity (raw)
Up to 4.59PB Raw (3.63 PB usable) per rack
Up to 2^128 objects
System Performance
Up to 10 GB/s bandwidth performance per rack (req's dual 40GbE connections to each rack TOR switch)
SSU Storage
Seagate 5U84 Storage enclosure, 84 drives, Integrated Storage Server, 10GbE Internal rack
network connections
SSU’s per rack
Up to 7 per rack (base or expansion)
Maximum Object Size
Unlimited (up to the size of the available storage)
Client Software
Seagate HSM Data Mover
Network Requirements
External system management interface (base rack only)
2x 1GbE
External data Interface
4x 40GbE (2x from each rack TOR switch)
External switch infrastructure
40GbE switch pair in a HA configuration with LACP support, VLAN support
A200 Disk Drives
Data Store
Seagate 8TB SMR SATA (Lamar) drives (82 per SSU)
Metadata Store
Seagate 6 Gb/s 6TB SAS drives (2 per SSU)
Dimensions
Height
1,991 mm (78.4 in)
Width
600 mm (23.62 in)
Depth
1,200 mm (47.24 in)
Weight
1,194Kg (2,26 lbs) (preliminary data)
System Availability
Data Availability
Four “nines” (99.99%) with 12 SSU configuration
(req's HA switch infrastructure and dual connections to each TOR switch)
Data Durability
Eleven “nines” (99.999%) with 12 SSU configuration
Hot Swappable Components
Disk Drives, Power Supplies, Fans, Power Cooling Modules and Server Modules
Power
Redundant Power Supplies and Power Cooling Modules
Power Cooling Modules
Dual fans per SSU
Power Consumption
Base Storage Rack (7 SSUs)
8.25 Kilowatts Peak (preliminary data)
Expansion Storage Rack (7 SSUs)
8.00 Kilowatts Peak (preliminary data)
Single SSU
1.10 Kilowatts Peak (preliminary data)
Heat Dissipation
28,150 BTU/hr
Base Storage Rack Configuration (7 SSUs)
ClusterStor HSM Specifications - General Information
Supported Target Systems
ClusterStor L - Series or third party Lustre 2.5.x system
Supported Archives
CS A200 archive
HSM PE/DB appliance
2U24 storage appliance including Seagate ClusterStor HSM SW (RobinHood)
HSM Data Mover software
Requires Customer Supplied Linux Server
Seagate-supplied Data mover 'copytool' software
1
Min 2 servers required, maximum supported 16
Network Requirements
2x 1GbE management interface
HSM PE/DB appliance
1x 40GbE/ FDR IB (connected to CS9000)
Customer Supplied Linux Server (for HSM Data Mover)
1x 40GbE (connected to A200) and
1x IB/40GbE (connected to CS9000)
Performance
IO Throughput
Up to 1.5 GB/s per HSM Data Mover
Update tracking method
Lustre RobinHood ChangeLog
2
Power Consumption
HSM PE/DB appliance
700W peak (preliminary data)
Heat Dissipation
HSM PE/DB appliance
2,388 BTU/hr
Note
1. HSM Data Mover Server Specifications
a.
Intel CPU (2609 v2 or greater), 32GB DRAM, 100GB Storage (OS).
b.
Centos 7.0 (7.0.1406) or compatible OS.
c.
Lustre 2.5.1 or later client software.
d.
40GbE ethernet HBA connected to the A200 network infrastructure.
e.
Network connection to Lustre network (IB/40GbE Ethernet).
f.
Ability to add CS-A200 specific software RPMs, including userspace applications and kernel modules.
2. Target reference platform requirements
a.
A200 archive configured with 12 SSU's and dual 40GbE connections to each A200
TOR switch.
b.
40GbE connection from data mover server to A200 network infrastructure.
seagate.com
AMERICAS
ASIA/PACIFIC
EUROPE, MIDDLE EAST AND AFRICA
Seagate Technology LLC 10200 South De Anza Boulevard, Cupertino, California 95014, United States, 408-658-1000
Seagate Singapore International Headquarters Pte. Ltd. 7000 Ang Mo Kio Avenue 5, Singapore 569877, 65-6485-3888
Seagate Technology SAS 16–18, rue du Dôme, 92100 Boulogne-Billancourt, France, 33 1-4186 10 00
© 2016 Seagate Technology LLC. All rights reserved. Printed in USA. Seagate, Seagate Technology and the Wave logo are registered trademarks of Seagate Technology LLC in the United States and/or other countries. ClusterStor is either a trademark or registered trademark of Seagate
Technology LLC or one of its affiliated companies in the United States and/or other countries. All other trademarks or registered trademarks are the property of their respective owners. When referring to drive capacity, one gigabyte, or GB, equals one billion bytes and one terabyte, or TB, equals
one trillion bytes. Your computer's operating system may use a different standard of measurement and report a lower capacity. In addition, some of the listed capacity is used for formatting and other functions, and thus will not be available for data storage. Actual data rates may vary depending on
operating environment and other factors. Seagate reserves the right to change, without notice, product offerings or specifications. CSES-DS133.1 1507US July 2015